Am'oSt"rimarm'}  Impurities  in  Water.  515 
subsequent  thawing  of  the  ice,  complete  separation  of  the  anti- 
mony sulphide  took  place.  When  a  very  large  quantity  of  the 
same  solution  was  exposed  to  a  low  temperature,  the  separation 
was  not  quite  complete.  A  colloidal  solution  of  copper  sulphide 
was  obtained  by  the  action  of  hydrogen  sulphide  on  copper  gly- 
collate.  The  separation  of  the  sulphide  was  incomplete,  although 
the  solution  was  kept  one  day  at  —  6°  and  one  day  at  —  190. 
Nothing  separated  out  at  the  ordinary  temperature.  Colloidal 
solutions  of  ferric  hydroxide,  obtained  by  the  dialysis  of  different 
preparations  of  basic  chloride,  sometimes  remain  uncoagulated 
by  freezing,  sometimes  are  partly  coagulated.  This  depends  on 
the  purity  of  such  solutions,  and  partly,  perhaps  on  the  rate  of 
cooling.  Solutions  of  potassium  ferric  tartrate  behaved  in  a  simi- 
lar manner.  One  part  of  starch  was  dissolved  in  100  parts  of  boil- 
ing water,  and  the  solution  was  kept  at  a  temperature  below  zero ; 
flocks  separated,  as  shown  already  by  Payen,  but  the  separation  was 
incomplete.  It  was  found  that  starch  solutions  which  had  been 
subjected  to  freezing  are  much  more  easily  and  quickly  filtered 
than  the  same  solutions  not  frozen,  so  that  this  process  might  be 
used  for  the  better  filtration  of  slimy  liquids  and  precipitates. 
Neither  white  of  egg  nor  milk  is  coagulated,  even  on  freezing  at 
—  1 5°  to  —  200.  The  fact  that  peat  falls  to  pieces  after  being 
frozen  is  explained  by  the  author  on  assumption  that  it  contains 
colloidal  substances,  which  are  coagulated  by  freezing. 
THE  EFFECT  OF  FREEZING  UPON  IMPURITIES 
CONTAINED  IN  WATER. 
The  Massachusetts  State  Board  of  Health  (June,  1890,)  publish 
experiments  with  seventy-six  samples  of  water,  and  336  samples  of 
ice  from  fifty-eight  localities  to  answer  the  above  question. 
In  ice  from  polluted  sources  compared  with  water  from  the  same, 
the  experiments  showed  : 
(1)  That  in  the  ice  the  color  and  salt  had  been  removed. 
(2)  That  all  but  13  per  cent,  of  the  other  impurities  of  the 
water,  as  shown  by  chemical  analysis,  had  been  removed. 
(3)  The  number  of  bacteria  in  the  cubic  centimetre  were :  For 
snow  (one  sample),  1,246;  for  clear  ice  (part  of  the  same  cake  as 
above),  6  ;  for  clear  ice  from  an  unpolluted  source,  o. 
(4)  The  average  of  12  samples  from  most  polluted  sources,  138 
